net.sourceforge.jiu.color.quantization

Class MedianCutQuantizer

Implemented Interfaces:
RGBIndex, RGBQuantizer

public class MedianCutQuantizer
extends ImageToImageOperation
implements RGBIndex, RGBQuantizer

Performs the Median Cut color quantization algorithm for a given list of colors.

Supported image types

The input image must implement RGB24Image.

Usage example

The following code snippet uses the default settings with a palette of 256 entries.
 MedianCutQuantizer quantizer = new MedianCutQuantizer();
 quantizer.setInputImage(image);
 quantizer.setPaletteSize(256);
 quantizer.process();
 PixelImage quantizedImage = quantizer.getOutputImage();
 
If you want to combine Median Cut quantization with error diffusion dithering to improve the visual quality of the output, try the ErrorDiffusionDithering class. However, note that noise is introduced into the image with dithering methods so that the resulting image may not be suitable for automatic processing.

Credits

The Median Cut algorithm was designed by Paul Heckbert. He described it in the article Color image quantization for frame buffer display. Comput. Graphics 16(3), 1982, 297 - 304. CiteSeer page of the article.
Author:
Marco Schmidt
See Also:
MedianCutContourRemoval, ErrorDiffusionDithering

Field Summary

static int
DEFAULT_METHOD_REPR_COLOR
The default method to determine the representative color from a list of colors.
static int
METHOD_REPR_COLOR_AVERAGE
Constant value for a method of determining the representative color for a set of colors by computing the average of all samples for each of the three components red, green and blue.
static int
METHOD_REPR_COLOR_MEDIAN
Constant value for a method of determining the representative color for a set of colors by picking the median value of all samples for each of the three components red, green and blue.
static int
METHOD_REPR_COLOR_WEIGHTED_AVERAGE
Constant value for a method of determining the representative color for a set of colors by computing the weighted average of all samples for each of the three components red, green and blue.
private boolean
doNotMap
private RGBColorList
list
private int
maxValue
private int
method
private boolean
outputTruecolor
private int
paletteSize
private MedianCutNode
root

Fields inherited from class net.sourceforge.jiu.ops.ImageToImageOperation

canInAndOutBeEqual, inputImage, outputImage

Fields inherited from class net.sourceforge.jiu.ops.Operation

abort, progressListeners

Fields inherited from interface net.sourceforge.jiu.data.RGBIndex

INDEX_BLUE, INDEX_GREEN, INDEX_RED

Constructor Summary

MedianCutQuantizer()
Creates a MedianCutQuantizer object and initializes its fields to default values.

Method Summary

private void
addNodes(MedianCutNode[] nodeList, MedianCutNode node)
private RGBColorList
createColorList(RGB24Image image)
MedianCutNode[]
createLeafList()
Creates a linear list of leaf nodes.
Palette
createPalette()
Creates a palette with the representative colors of all leaf nodes.
void
findAllRepresentativeColors()
private MedianCutNode
findLeafToBeSplit(MedianCutNode node)
Traverses tree given by argument node and returns leaf with largest distribution of samples for any of its three components.
MedianCutNode
findNearestNeighbor(int[] rgb)
For a given RGB value, searches the node in the internal node tree whose representative color is closest to this color.
int
findNearestNeighbor(MedianCutNode[] nodes, int red, int green, int blue)
For each node in the argument array computes the distance between the representative color of that node and the color given by the three argument samples.
void
findPalette()
private int[]
findRepresentativeColor(int index1, int index2)
Computes a representative color for a set of colors in the color list.
private void
findRepresentativeColors(MedianCutNode node)
Calls findRepresentativeColor with node if node is a leaf.
int
getMethodToDetermineRepresentativeColors()
Returns the method (to be) used to determine the representative color for the list of colors of a node.
int
getPaletteSize()
Returns the number of colors in the destination image.
boolean
getTruecolorOutput()
Returns if this operation is supposed to generate truecolor or paletted output.
int
map(int[] origRgb, int[] quantizedRgb)
This method maps a triplet of intensity values to its quantized counterpart and returns the palette index of that quantized color.
void
mapImage(RGB24Image in, Paletted8Image out)
void
mapImage(RGB24Image in, RGB24Image out)
void
process()
This method does the actual work of the operation.
void
setAllPaletteIndexValues()
void
setMapping(boolean doMap)
Defines whether process will map the input image to an output image.
void
setMethodToDetermineRepresentativeColors(int newMethod)
Sets the method to determine the representative color for a list of colors.
private int
setPaletteIndexValues(MedianCutNode node, int index)
Recursively visits node and its descendants, assigning ascending palette index values to leaves via MedianCutNode.setPaletteIndex(int).
void
setPaletteSize(int newPaletteSize)
Sets the number of colors that this operations is supposed to reduce the original image to.
void
setTruecolorOutput(boolean useTruecolor)
Lets the user specify if the output image is to be truecolor (argument useTruecolor is true) or paletted (argument useTruecolor is false).
void
splitNode(MedianCutNode node)

Methods inherited from class net.sourceforge.jiu.ops.ImageToImageOperation

canInputAndOutputBeEqual, ensureImagesHaveSameResolution, ensureInputImageIsAvailable, ensureOutputImageResolution, getInputImage, getOutputImage, setCanInputAndOutputBeEqual, setInputImage, setOutputImage

Methods inherited from class net.sourceforge.jiu.ops.Operation

addProgressListener, addProgressListeners, getAbort, process, removeProgressListener, setAbort, setProgress, setProgress

Field Details

DEFAULT_METHOD_REPR_COLOR

public static final int DEFAULT_METHOD_REPR_COLOR
Field Value:
2

METHOD_REPR_COLOR_AVERAGE

public static final int METHOD_REPR_COLOR_AVERAGE
Constant value for a method of determining the representative color for a set of colors by computing the average of all samples for each of the three components red, green and blue. #getMethodToDetermineRepresentativeColors #setMethodToDetermineRepresentativeColors
Field Value:
0

METHOD_REPR_COLOR_MEDIAN

public static final int METHOD_REPR_COLOR_MEDIAN
Constant value for a method of determining the representative color for a set of colors by picking the median value of all samples for each of the three components red, green and blue. #getMethodToDetermineRepresentativeColors #setMethodToDetermineRepresentativeColors
Field Value:
2

METHOD_REPR_COLOR_WEIGHTED_AVERAGE

public static final int METHOD_REPR_COLOR_WEIGHTED_AVERAGE
Constant value for a method of determining the representative color for a set of colors by computing the weighted average of all samples for each of the three components red, green and blue. Weighted means that each color is multiplied by the number of times it occurs in the input image. The values of samples multiplied by their frequency are then divided by the total number of times the colors appear in the image. #getMethodToDetermineRepresentativeColors #setMethodToDetermineRepresentativeColors
Field Value:
1

doNotMap

private boolean doNotMap

list

private RGBColorList list

maxValue

private int maxValue

method

private int method

outputTruecolor

private boolean outputTruecolor

paletteSize

private int paletteSize

root

private MedianCutNode root

Constructor Details

MedianCutQuantizer

public MedianCutQuantizer()
Creates a MedianCutQuantizer object and initializes its fields to default values.

Method Details

addNodes

private void addNodes(MedianCutNode[] nodeList,
                      MedianCutNode node)

createColorList

private RGBColorList createColorList(RGB24Image image)
            throws OperationFailedException

createLeafList

public MedianCutNode[] createLeafList()
Creates a linear list of leaf nodes. Assumes that findPalette() was successfully run before.

createPalette

public Palette createPalette()
Creates a palette with the representative colors of all leaf nodes. Assumes that findPalette() was successfully run before.
Specified by:
createPalette in interface RGBQuantizer
Returns:
palette with all representative colors

findAllRepresentativeColors

public void findAllRepresentativeColors()

findLeafToBeSplit

private MedianCutNode findLeafToBeSplit(MedianCutNode node)
Traverses tree given by argument node and returns leaf with largest distribution of samples for any of its three components.

findNearestNeighbor

public MedianCutNode findNearestNeighbor(int[] rgb)
For a given RGB value, searches the node in the internal node tree whose representative color is closest to this color.
Parameters:
rgb - the color for which a match is searched; the array must have at least three entries; RGBIndex constants are used to address the samples
Returns:
node with best match

findNearestNeighbor

public int findNearestNeighbor(MedianCutNode[] nodes,
                               int red,
                               int green,
                               int blue)
For each node in the argument array computes the distance between the representative color of that node and the color given by the three argument samples.
Returns:
index of the node with the smallest distance or -1 if the array has a length of 0

findPalette

public void findPalette()

findRepresentativeColor

private int[] findRepresentativeColor(int index1,
                                      int index2)

findRepresentativeColors

private void findRepresentativeColors(MedianCutNode node)
Calls findRepresentativeColor with node if node is a leaf. Otherwise, recursively calls itself with both successor nodes.

getMethodToDetermineRepresentativeColors

public int getMethodToDetermineRepresentativeColors()
Returns:
the method, one of the METHOD_xyz constants

getPaletteSize

public int getPaletteSize()
Returns the number of colors in the destination image. If output is paletted, this is also the number of entries in the palette.
Returns:
number of colors in the destination

getTruecolorOutput

public boolean getTruecolorOutput()
Returns if this operation is supposed to generate truecolor or paletted output.
Returns:
if truecolor images are to be generated

map

public int map(int[] origRgb,
               int[] quantizedRgb)
This method maps a triplet of intensity values to its quantized counterpart and returns the palette index of that quantized color. The index values for the two arrays are taken from RGBIndex.
Specified by:
map in interface RGBQuantizer
Parameters:
origRgb - the three samples red, green and blue for which a good match is searched in the palette
quantizedRgb - will hold the three samples found to be closest to origRgb after the call to this method
Returns:
int index in the palette of the match quantizedRgb

mapImage

public void mapImage(RGB24Image in,
                     Paletted8Image out)

mapImage

public void mapImage(RGB24Image in,
                     RGB24Image out)

process

public void process()
            throws MissingParameterException,
                   OperationFailedException,
                   WrongParameterException
This method does the actual work of the operation. It must be called after all parameters have been given to the operation object.
Overrides:
process in interface Operation
Throws:
WrongParameterException - if at least one of the input parameters was not initialized appropriately (values out of the valid interval, etc.)
MissingParameterException - if any mandatory parameter was not given to the operation
OperationFailedException -

setAllPaletteIndexValues

public void setAllPaletteIndexValues()

setMapping

public void setMapping(boolean doMap)
Defines whether process will map the input image to an output image. If not, only the palette is determined.

setMethodToDetermineRepresentativeColors

public void setMethodToDetermineRepresentativeColors(int newMethod)
Sets the method to determine the representative color for a list of colors. After the algorithm has determined sets of colors that lie closely together in color space and will be mapped to the same color in the destination image, the algorithm will determine that color
Parameters:
newMethod - the new method, one of the METHOD_xyz constants in this class

setPaletteIndexValues

private int setPaletteIndexValues(MedianCutNode node,
                                  int index)
Recursively visits node and its descendants, assigning ascending palette index values to leaves via MedianCutNode.setPaletteIndex(int). If this method is called with root and 0 as parameters, all leaves will get a unique palette index.

setPaletteSize

public void setPaletteSize(int newPaletteSize)
Sets the number of colors that this operations is supposed to reduce the original image to.
Parameters:
newPaletteSize - the number of colors

setTruecolorOutput

public void setTruecolorOutput(boolean useTruecolor)
Lets the user specify if the output image is to be truecolor (argument useTruecolor is true) or paletted (argument useTruecolor is false). If the color type is to be changed afterwards, use PromoteToRgb24 to convert from paletted to truecolor. Reducing a truecolor image that uses only 256 or less colors to a paletted image can be done with AutoDetectColorType.
Parameters:
useTruecolor -

splitNode

public void splitNode(MedianCutNode node)